State Attorneys General Sue OpenAI Under Consumer-Protection Law Over ChatGPT Safety Claims

What did Iowa and Idaho file?

Each filed a near-parallel pleading on October 14, 2026, against the same eight OpenAI entities. Iowa filed a petition in the Iowa District Court for Polk County. Idaho filed a complaint in the Fourth Judicial District Court, Ada County. The copies posted by both offices show a blank case number, so we give no docket for either.

Iowa pleads two causes of action under the Iowa Consumer Fraud Act, Iowa Code 714.16. The first alleges unfair and deceptive practices in the marketing and design of ChatGPT. Iowa alleges, among other things, false assurances that ChatGPT is safe, a privacy-policy statement that OpenAI does not collect personal data from children or minors, presentation of ChatGPT as having human characteristics through first-person pronouns and emotive language, and a "13+" Apple App Store rating. The second cause of action alleges unfair and deceptive practices in testing autonomous AI agents, including failing to disclose incidents in which models acted outside their testing environment. Iowa asks for a civil penalty of "up to $40,000 for each violation of the Iowa Consumer Fraud Act" and up to $5,000 more per violation against an older individual. It also asks for restitution, disgorgement, deletion of Iowans' data, and injunctions, including one against further testing of advanced models until safeguards and third-party monitors are in place.

Idaho pleads two counts under the Idaho Consumer Protection Act, I.C. 48-601 et seq., on closely matching facts: Count I on ChatGPT and Count II on testing of advanced AI models. Its prayer asks the court to order "a civil penalty of $5,000 for each violation of the Idaho Consumer Protection Act," plus damages or restitution, disgorgement and injunctions. The injunctions Idaho requests include barring deployment or testing of advanced models outside a secured, independently verified environment, requiring prompt disclosure to the attorney general of any incident in which a model escapes its controls, and additional measures such as age verification and limits on harmful or addictive features.

What do we know about the Arkansas complaint?

We know its court and docket for certain, and we read its body by machine. The clerk's stamp shows Faulkner County Circuit Court, case number 23CV-26-2984, e-filed October 14, 2026 at 10:36:32, 73 pages. The PDF is a scanned image with no text layer, so we ran optical character recognition across all 73 pages. We read the introduction, the statutory counts, the request for relief and the signature block from that output. We did not compare every passage against the page images, so we quote it sparingly and treat details as unconfirmed against the image.

On that reading, Arkansas alleges four counts under the Arkansas Deceptive Trade Practices Act, three counts under the Arkansas Children and Teens Online Privacy Protection Act, and five common-law counts: negligence, strict liability for design defect, strict liability for failure to warn, unjust enrichment and public nuisance. It requests civil penalties of $10,000 per violation, with an additional $10,000 per violation against an elder person or person with a disability, and it demands a jury trial.

What do the other attorneys general say?

The Iowa attorney general's release said a bipartisan group of states "have also individually sued OpenAI," that the Kansas, Nebraska and West Virginia attorneys general joined her press conference, and that "Arkansas, Arizona, Idaho and Montana have also sued OpenAI in their own state courts." The West Virginia attorney general said he filed a lawsuit the same day under West Virginia consumer protection laws. He did not state a court in the release. The Arizona attorney general said her complaint was "filed today in Maricopa County Superior Court" and alleges violations of the Arizona Consumer Fraud Act.

We did not open the Arizona or West Virginia complaints, and we did not open any Kansas, Montana or Nebraska document. Beyond the attributed statements above about Kansas, Nebraska and Montana, we state no court, docket, statute or claim for those three states.

How are the three pleadings related, and how do they differ from the Florida suit?

They name the same eight defendants and list the same outside counsel, Kibler Fowler & Cave LLP. Those two shared features are the only links between them that we verified. The releases describe separate suits in each state's own court, and we found no joint complaint. The statutes differ, and so do the requested penalties and the relief.

The Florida attorney general sued OpenAI on June 1, 2026, in a different court, and that complaint also named Sam Altman personally. We cover it in Florida Sues OpenAI and Sam Altman Personally Over ChatGPT. The October 14 filings are a later event by other states, and the Iowa, Idaho and Arkansas pleadings we read name corporate defendants only.

What does it mean for general counsel, product and compliance leads, and state-regulatory practitioners?

It adds no duty, but it shows the theory three states chose: that public statements about a chatbot's safety, age-appropriateness and human-likeness can be deceptive or unfair under existing consumer law. The pleadings put specific statements in play, including a general safety commitment, an app-store age rating, a privacy-policy sentence about minors and first-person presentation. Iowa also pleads product design, safety testing and warnings as unfair practices, alongside advertising claims. It further alleges that the evidence OpenAI held "did not provide a reasonable basis" for its representations.

On counting, the Iowa petition alleges that "Each of Defendants' misrepresentations is a separate violation," which is how per-violation penalties would multiply if a court agreed. Penalty ceilings and statutory text differ across the three states, so a company active in all three faces different exposure in each.

What should a company with a consumer chatbot do now?

Start with an inventory, because the pleadings turn on specific statements. This is our reading of the theories pleaded, not a requirement stated by any court.

• List public statements about safety, reliability, age-appropriateness and human-like behavior across your site, app listings, terms and privacy policy, and attach the evidence for each.
• Check that your app-store age rating, your privacy policy's statements about minors and your actual age-assurance practice describe the same thing.
• Assign ownership for disclosing incidents where a model or agent acts outside its intended limits.
• If you serve users in Iowa, Idaho or Arkansas, read the statute each state pleaded.
• Watch the dockets. We saw no ruling, and the Iowa and Idaho case numbers were blank in the copies we opened.

What we did not verify

What we opened: the full Iowa petition and the full Idaho complaint (both with text layers), the Arkansas complaint (scanned; read by optical character recognition, not compared page by page against the images), and the Iowa, Arizona and West Virginia attorney general releases. Quotations from the Iowa and Idaho pleadings are from their text layers with typographic quotes folded to straight ones.

What we did not open: the Arizona complaint, the West Virginia complaint, and any Kansas, Montana or Nebraska document. Statements about those three states on this page are the Iowa and West Virginia attorneys general's own, attributed to them. We did not check the Polk County, Ada County or Faulkner County dockets for later filings, so current procedural status is unverified. We found no OpenAI response and no removal to federal court in the sources we opened.

What we refuse to claim: that any allegation is true, including the pleadings' allegations about suicide, self-harm, violence, deaths and model behavior; how any case will end; that the suits are coordinated beyond the shared defendants and shared counsel noted above; or that eight state courts have eight verified dockets.

Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.

Related briefings

Up to the AI Regulation News hub, the North America AI regulation pillar and the AI disclosure rules topic hub.

• Florida Sues OpenAI and Sam Altman Personally Over ChatGPT
• Texas Attorney General's TRAIGA Enforcement Powers Are Now Live
• 26 Attorneys General Ask Congress for Frontier AI Oversight and No Preemption
• California SB 243: Companion-Chatbot Safety Duties